Loss of IVNS1ABP, a gigaxonin paralogue, leads to a progeroid neuropathy due to impaired proteostasis

A homozygous variant in IVNS1ABP segregates with a progeroid neuropathy

We investigated a consanguineous family in which three affected patients exhibit an early-onset and progressive neuropathy (Table 1 and Figures 1A and 1B). The proposita II.1, II.4 and II.5 display peripheral neuropathy leading to steppage gait with progressive loss of deambulation (Figure S1A). Patient II.1 walked at 1 year old, but showed signs of regression at 5-10 years old, with difficulty climbing the stairs at 5-10 years old, and lost gait at the age of 10-15. Patients II.4 and II.5 walked at 1.5 years old and showed abnormal gait from 5-10 years old (Movie 1). Skin dyschromatosis with hypo- and hyper-pigmented macules over the face and body were present, the extent of which increased with age (Figure S1B). A notable premature whitening of hairs, eyelashes and eyebrows was seen in all three affected patients (Figure 1B). Consistent with autosomal recessive inheritance, homozygosity mapping followed by exome sequencing in patient II.1 identified a biallelic germline missense variant in the IVNS1ABP gene (MIM609209) on chr1:185274675 (GRCh37/hg19) (Table 2 and Figures 1C and S1C). The private mutation NM_006469.4 c.758T>G, p.(Phe253Cys) fully segregated with the disease, the three patients being homozygous, and the unaffected family members either non-carrier (II.2) or heterozygous (II.3, II.6) (Figure 1A). According to publicly available databases and bioinformatics tools, this private allele p.F253C has never been reported in the population and is predicted to be pathogenic (SIFT=0.01, PolyPhen-2=0.993, M-CAP=0.022) (Table 2). According to gnomAD v4.1, no homozygous damaging variants have been reported for IVNS1ABP, and the pLoF observed/expected score is 0.32, suggesting that IVNS1ABP is intolerant to loss-of-function mutations. A high Combined Annotation Dependent Depletion (CADD) score of 29.9 further demonstrates that this homozygous variant may be deleterious (Figure 1D).

• Download figure
• Open in new tab

Figure 1.

A recessive p.(Phe253Cys) IVNS1ABP variant segregates with a segmental syndrome of premature aging

(A) Pedigree of a consanguineous family with three affected patients carrying a homozygous p.(Phe253Cys) IVNS1ABP variant. Black fill, affected individual; white fill, unaffected individual; green asterisk, skin biopsy; red asterisk, blood drawn. Unaffected family members are either heterozygous (F253C/WT) or non-carrier (WT/WT). Skin biopsies were obtained from unaffected family members (heterozygous I.1, non-carrier II.2) and two affected patients (II.1, II.4).

(B) Photographs of the three affected patients displaying premature hair whitening and skin dyschromatosis, long fingers, club feet and steppage gait. See also Figures S1A and S1B.

(C) Homozygosity mapping narrowed down four identical-by-descent (IBD) blocks, and exome sequencing delineated 1 deleterious homozygous (Hmz) variant in IVNS1ABP. See also Figure S1C.

(D) Minor allele frequency and combined annotation dependent depletion (CADD) score of homozygous IVNS1ABP coding variants found in gnomAD v.2.1.1 (black dots) and those found in the affected patients (red dot).

(E) Domain schematic of the IVNS1ABP protein predicted by Alphafold3, with an N-terminal BTB domain and a C-terminal Kelch domain. In between, a truncated BACK domain with just its 3-box, a ubiquitin-like β-grasp domain, and two intrinsically disordered regions (IDR1 and IDR2), one within the β-grasp and one preceding the Kelch domain. The mutated residue p.F253 (located in IDR1) is highly conserved throughout vertebrate evolution.

(F) 3D conformation of IVNS1ABP homodimer predicted by Alphafold3.

(G) Co-immunoprecipitation of transiently overexpressed Ha-tagged wildtype IVNS1ABP^WT or mutant IVNS1ABP^F253C by Flag-tagged KLHL20, a known IVNS1ABP interacting protein.

(H) AlphaFold3 prediction of the binding of IVNS1ABP β-grasp fold to KLHL20 Klech domain. See also Figure S1D.

IVNS1ABP encodes the Influenza Virus NS1A-Binding Protein IVNS1ABP (Q9Y6Y0), also known as NS1-BP or KLHL39, a member of the BTB-Kelch family. Proteins in this family typically contain an N-terminal BTB domain and a C-terminal Kelch domain separated by an intervening linker region known as the BACK domain^32–34. X-ray crystal structures have been determined for the individual BTB and Kelch domains of IVNS1ABP^35,36, but not for the full-length protein. The structural models of IVNS1ABP predicted by AlphaFold3³⁷ suggested a truncated BACK domain with only its 3-box motif which comprises the first two α-helices. Intriguingly, the remaining region is replaced by a ubiquitin-like β-grasp fold and two intrinsically disordered regions (IDR1 and IDR2), a likely unique feature for this family (Figure 1E). Phenylalanine 253 is located in IDR1, which forms a 51-residue disordered loop within the ubiquitin-like β-grasp domain (Figures 1E and 1F). IDRs often form sites of protein-protein interaction and can undergo disorder-to-order transitions upon complexing with their partners. Mutations in many proteins with IDRs have been implicated in human diseases, such as neurodegeneration³⁸. Importantly, Phe253 is conserved across IVNS1ABP orthologues, suggesting that its alteration may disrupt IVNS1ABP-protein interactions.

To examine the consequence of the IVNS1ABP missense variant p.F253C on the binding to one of its direct interacting partners KLHL20³⁹, overexpression of IVNS1ABP full-length and KLHL20 Kelch domain in HEK293T cells followed by co-immunoprecipitation was carried out. Unlike wildtype (WT) HA-tagged IVNS1ABP^WT, the mutant IVNS1ABP^F253C could no longer be pulled down by Flag-tagged KLHL20, suggesting that the p.F253C variant may introduce a conformational change in its native folding affecting KLHL20 engagement (Figure 1G). Notably, AlphaFold3 predicted that IVNS1ABP Phe253 forms hydrophobic interactions to KLHL20 in a similar manner as DAPK1 Leu1339 binds to KLHL20⁴⁰ (Figure 1H and S1D). In agreement, previous immunoprecipitation experiments mapped IVNS1ABP residues 234-350, encompassing all of IDR1 and IDR2, as an essential region for many of the native interactions of IVNS1ABP³⁵.

Altogether, these results indicate that the conserved region around the disease mutation site is required for IVNS1ABP interaction with its targets, supporting the deleterious effect of the p.F253C variant on IVNS1ABP function.

Past phylogenetic analyses performed on 38 BTB-Kelch family proteins⁴¹ showed that IVNS1ABP is a close paralogue of the E3 ligase adaptor gigaxonin. The genetic inactivation of gigaxonin (also known as KLHL16, MIM605379) is responsible for Giant Axonal Neuropathy-1 (GAN1, MIM256850), a severe neurodegenerative disease⁴². The comparison of GAN1’s symptomatology with the clinical presentations of this novel syndrome revealed a strong overlap, including a similar progressive peripheral neuropathy leading to muscle weakness, hypotonia and steppage gait. Skin hyperpigmentation, scoliosis, long fingers, club feet, pes cavus, distal sensory impairment, dysarthria, reflex dysfunction, mental retardation were all shared between the two diseases (Table 1, Figures 1B and S1D). One of the distinguishing but related phenotypes is that GAN1 patients often present with kinky red hair whereas the three affected patients herein displayed premature hair whitening. The clinical similarities and protein sequence homology between IVNS1ABP and gigaxonin suggested that both proteins may fulfill similar functions by maintaining cellular proteostasis.

IVNS1ABP^F253C induces protein aggregation in patients’ cells

To examine the pathogenicity of this private IVNS1ABP^F253C variant, primary fibroblasts isolated from skin biopsies of 2 affected patients (II.1, II.4), a healthy ‘non-carrier’ (II.2) and heterozygous (I.2) family members were reprogrammed into induced pluripotent stem cells (iPSCs). These iPSC lines were then differentiated into disease-relevant cell lineages including neural progenitor cells (NPCs) and post-mitotic neurons (Figure 2A). Using an aggresome detection assay, patients’ primary fibroblasts (IVNS1ABP^F253C/F253C) showed significant increase of protein aggregates relative to WT cells (IVNS1ABP^WT/WT) (Figure 2B). This was also observed in the patient’s iPSCs and iPSC-derived neurons (Figures 2C, 2D and S2A). These results suggest that IVNS1ABP^F253C prevents proper degradation of endogenous proteins leading to their aggregation across many cell types. To assess the overall cellular proteostasis, we next examined autophagy lysosomal activity which drives the degradation and recycling of cellular wastes. Immunostaining of the lysosome-associated membrane proteins 1 and 2 (LAMP1/2) showed significant enlargement of the lysosomes in each patient cell type studied (Figures 2E and S2B). Increased level of P62, a classical marker of aggrephagy⁴³, was also observed by western blot in IVNS1ABP^F253C/F253C compared to IVNS1ABP^WT/WT and IVNS1ABP^WT/F253C NPCs (Figure 2F). These results indicate that maintenance of cellular proteostasis in patients’ cells is significantly altered, causing protein aggregation and overactivation of the autophagy lysosomal pathway.

• Download figure
• Open in new tab

Figure 2.

Altered IVNS1ABP induces protein aggregation in patients’ primary cells

(A) Schematic representation of the different cells generated from punch skin biopsy of patients and controls, from primary fibroblasts, to induced pluripotent stem cells (iPSCs), to neural progenitor cells (NPCs), to differentiated neurons.

(B) Proteostat® staining and quantification show significant increase of protein aggregates in patient’s primary fibroblasts (II.4) (n=31) compared to unaffected non-carrier cells (II.2) (n=19). Scale bar = 50 µm. Error bars indicate mean ± SEM. Unpaired t test, *** p < 0.001.

(C) Proteostat® staining and quantification show significant increase of protein aggregates in patient’s iPSCs (II.1) compared to unaffected non-carrier iPSCs (II.2) (each dot represents one protein aggregate. Data were collected from n≥3 independent cultures). Scale bar = 20 µm. Error bars indicate mean ± SEM. Unpaired t test, **** p < 0.0001. See also Figure S2A.

(D) Proteostat® staining and quantification show increase of protein aggregates in patient’s iPSC- derived neurons (II.1) compared to unaffected non-carrier cells (F.II.2) (each dot represents one culture condition. Data were collected from n=3 independent cultures). iPSC-derived neurons expressed Microtubule-associated protein 2 (MAP2). Scale bar = 50 µm. Error bars indicate mean ± SEM. Unpaired t test, ns: non-significant.

(E) Immunostaining and quantification of lysosome-associated membrane protein 2 (LAMP2) show significant increase of the lysosomal network in patient’s primary fibroblasts (II.4), iPSCs (II.1) and patient’s iPSC-derived neurons (II.1) compared to unaffected cells (II.2) (One dot represents one lysosome, data were collected from n≥3 independent cultures in each group). Scale bar = 20 µm. Error bars indicate mean ± SEM. Unpaired t test, **** p < 0.0001. See also Figure S2B.

(F) Western blot of patient’s NPCs (II.4) treated with MG132 shows increased P62 (autophagy marker) compared to unaffected family members (I.1) and (II.2). GAPDH was used as a loading control.

Loss of IVNS1ABP, or gigaxonin, result in a similar accumulation of ubiquitinated substrates

IVNS1ABP belongs to the BTB-Kelch protein family, members of which serve as adapters for the ubiquitination of protein substrates by multi-subunit E3-ubiquitin ligase complexes. Based on IVNS1ABP’s homology to gigaxonin, we next examined its role in the ubiquitination of protein substrates^33,39,44,45. Upon cell starvation and Epoxomicin-mediated inhibition of the ubiquitin proteasome pathway (UPP), we found that, compared to WT NPCs, mutant IVNS1ABP^F253C/F253C cells displayed evidence of accumulation of ubiquitinated proteins (Figure 3A). To obtain a quantitative measure of these differences and identify possible substrates of IVNS1ABP, we employed targeted mass spectrometry following ubiquitin-mediated enrichment from whole cell extracts (Figure 3B). Significant differences in the total amount of ubiquitinated proteins from the two homozygous patients compared to WT or heterozygous NPCs could be readily quantified (Figures 3C). Remarkably, four of the top 10 differentially ubiquitinated proteins in patients’ NPCs are established substrates of gigaxonin, including Vimentin^46–49, Tubulin-binding cofactor B⁵⁰,

• Download figure
• Open in new tab

Figure 3.

Protein ubiquitination is altered in IVNS1ABP mutant cells

(A) Western blot of patient’s starved NPCs (II.4), in presence or absence of Epoxomicin, shows increased ubiquitinated proteins compared to unaffected control (II.2). GAPDH was used as a loading control.

(B) Schematic representation of mass spectrometry workflow to quantify ubiquitinated proteins in whole cell lysates.

(C) Quantification by mass spectrometry of whole-NPCs lysates from two patients (II.1 and II.4) shows significant increase of ubiquitinated peptides (normalized to total proteins)

compared to two healthy controls (I.1 and II.2). Y-axis represents log10 abundances of normalized ubiquitinated peptides. *** p-value<0.001.

(D) List of the top 10 differentially ubiquitinated proteins in two patients’ NPC extracts (II.1 and II.4) compared to two controls (I.1 and II.2). Five of these are known to be direct substrates of gigaxonin or gigaxonin’s targets. Proteins were ranked by the p-values in ascending order. See also Table S1 and Figure S3.

(E) Western blot of patient’s NPCs (F253C/F253C, II.4) shows increased endogenous MAP1B compared to whole-cell lysates of unaffected control (WT/WT, II.2), with stronger differences when cells were treated with Epoxomicin. MAP1B LC, Light Chain; FL, Full Length; HC, Heavy Chain. GAPDH serves as loading control.

(F) Western blot of patient’s primary fibroblasts (F253C/F253C, II.1) also shows increased endogenous MAP1B compared to whole-cell lysates of unaffected control (WT/WT, II.2).

(G) Western blot of endogenous IVNS1ABP, MAP1B and gigaxonin in immortalized fibroblasts of unaffected control II.2 (WT/WT, lane 1), IVNS1ABP-mutated patient II.1 (F253C/F253C, lane 2), GAN-null patient (-/-, lane 3), CRISPR-Cas9-induced IVNS1ABP knockout (-/-) in GAN-null patient (-/-, lane 4) and CRISPR-Cas9-induced IVNS1ABP knockout (-/-) in unaffected control II.2 (WT/WT, lane 5). All cell lines show increased MAP1B compared to unaffected WT/WT control II.2. MAP1B LC, Light Chain; FL, Full Length; HC, Heavy Chain. GAPDH was used as a loading control.

Peroxiredoxin-1⁴⁶ and 14-3-3 proteins⁵¹ (Figure 3D, Table S1 and Figure S3). Other hits included the ubiquitin-conjugating enzyme E2 L3 (UBE2L3) which was shown to bind MAP1B, a prototypical substrate of gigaxonin⁵² (Figure 3D). This prompted us to evaluate the levels of MAP1B in patients’ NPCs by western blot. We observed that, with or without UPP inhibition by Epoxomicin, MAP1B full length or its cleaved light and heavy chains, were noticeably increased in patient cells relative to heterozygous IVNS1ABP^F253C/WT cells (Figure 3E). The accumulation of MAP1B being a tell-tale sign of Gixagonin deficiency in GAN1 patient-derived cells, we sought to see if this was also true in patient’s primary fibroblasts. Compared to heterozygous fibroblasts, endogenous MAP1B polyprotein and subunits were robustly increased in IVNS1ABP^F253C/F253C cells (Figure 3F). These results support the notion that IVNS1ABP and gigaxonin earmark similar substrates for degradation, explaining why their genetic inactivation results in a similar set of neurodegenerative phenotypes.

To address whether Gixagonin and IVNS1ABP are functionally redundant, we sought to characterize single and double knockout fibroblasts for these two paralogous genes. First we derived fibroblasts from a GAN1 patient with a validated null homozygous allele which deletes the entirety of the GAN locus (breakpoints [GRCh37]16q23.2 [81262391_81445961]). The GAN-null patient showed similar musculoskeletal defects, progressive and severed neuropathy as did the three IVNS1ABP-deficient patients (Figure S1D and Table 1). By CRISPR-Cas9, we then knocked out IVNS1ABP in WT or GAN-null cells. These engineered and immortalized fibroblasts were confirmed to be single or double KO for IVNS1ABP and/or GAN by genomic sequencing (data not shown) and by western blotting (Figure 3G). Using endogenous MAP1B levels as a proxy for IVNS1ABP and gigaxonin activity, we observed that IVNS1ABP^F253C/F253C (lane 2) and IVNS1ABP^KO/KO (lane 5) cells equally result in the accumulation of MAP1B, indicating that this recessive missense allele behaves as KO allele. As expected, the loss of gigaxonin (lane 3) resulted in accrued MAP1B levels, which was comparable to that of IVNS1ABP KO cells, but was not further enhanced in GAN/IVNS1ABP double KO cells (lane 4) (Figure 3G). These unambiguous KO alleles and pairwise isogenic cell lines proved to be robust readouts for addressing the complementarity of gigaxonin and IVNS1ABP. We find that both E3 ligase adapter proteins are needed for the degradation of common cytoskeletal substrates such as MAP1B, the accumulation of which may lead to neurotoxicity.

Axonal damage is rescued in genetically-corrected IVNS1ABP motor neurons

To further prove the pathogenicity of this private IVNS1ABP variant and mitigate the effect of possible genetic modifiers, isogenic lines were created using CRISPR-Cas9. To this end, the homozygous mutation (c.758T>G; p.F253C) in patient’s iPSCs (II.1) was genetically and bi- allelically corrected into its WT genotype (c.758T/T)⁵³ (Figure 4A). Patient-derived and their isogenic WT iPSCs (WT^corr./WT^corr.) efficiently differentiated into motor neuron (MN) progenitors, expressing MN markers such as OLIG2 and MNX1⁵⁴ (Figure S4). After 4 weeks of culture, these paired isogenic MN progenitors differentiated into mature MNs expressing choline acetyltransferase (ChAT) (Figure 4B). Less abundant and enlarged neurite bundles extending from the MN spheres marked by MAP2 were observed in mutant IVNS1ABP^F253C/F253C cells compared to isogenetically-corrected WT ones (Figure 4B). As previously observed in non- isogenic fibroblasts, iPSCs and neurons (Figure 2E), enlarged lysosomes in IVNS1ABP^F253C/F253C MNs compared to isogenic WT were documented (Figure 4C). The use of a lysotracker and live imaging revealed slower intra-cellular movement of lysosomes which was rescued in IVNS1ABP^Corr./Corr. MNs (Figure 4D). Immunostaining of neural filaments (NF) using phosphorylated p-NF, an established marker of axonal damage⁵⁵, showed abnormal p-NF accumulation especially in the distal ends of mutated axons, suggesting potential axon transport deficit (Figure 4E). Western blot analysis confirmed a noticeable increase of p-NF triple proteins (heavy, medium and light chains) in patient’s MNs compared to WT corrected ones (Figure 4F).

• Download figure
• Open in new tab

Figure 4.

IVNS1ABP mutation leads to human motor neuron degeneration

(A) Mutation F253C/F253C in patient’s iPSCs (II.1) was corrected using CRISPR-Cas9, creating isogenic control iPSCs (WT^corr./WT^corr.). Spinal cord motor neurons (MNs) were differentiated from both isogenic iPSCs.

(B) Immunostaining of ChAT and MAP2 shows that isogenic motor neuron progenitors differentiated further into mature MNs.

(C) Lysotracker representative image and quantification show significant increase of the lysosome size in patient’s iPSC-derived MNs (F253C/F253C) compared to its isogenic wildtype cell line (WT^corr./WT^corr.). Error bars indicate mean ± SEM. Unpaired t test, ** p < 0.01 (n=5 culture conditions).

(D) Live imaging of lysotracker dye in iPSC-derived MNs (4-week-old) shows impaired intracellular lysosome movement over time in patient’s MNs (n=87) compared to its isogenic wildtype MNs (n=80). Each dot represents one organelle labeled by lysotracker. Data was collected across three independent cultures.

(E) Immunostaining with phospho-neurofilament antibodies (p-NF) shows that mutant F253C/F253C MNs display p-NF accumulation at the distal ends of MNs compared to isogenic WT^corr./WT^corr. cells. Each data point represents the ratio of damaged axons (labeled with pNF aggregates) in each image field taken from 3 independent MN cultures. Error bars indicate mean ± SEM. Unpaired t test, * p < 0.05.

(F) Western blot of patient MNs show increased p-NF compared to its isogenic control cells. p- NF-H, heavy chain; p-NF-M, medium chain; p-NF-L, light chain. NSE, a neuron-specific enolase, is used as a loading control.

p-NF is a pathological hallmark for axonal damage in multiple neurodegenerative diseases, such as amyotrophic lateral sclerosis, Alzheimer’s disease and GAN1^56,57. Altogether these in vitro experiments using isogenic pairs of iPSC-derived MNs confirmed the deleterious effect of p.F253C, which we speculate may be responsible for motor neuron degeneration in the three affected patients.

Ivns1abpa/b KO zebrafish show defective motor neuron development and locomotion

To better understand the physiological role of Ivns1abp, we next sought to create an animal model of the disease. We favored zebrafish over mice, following the observation that gan KO fish more faithfully recapitulated the severity of human disease than did mouse models of GAN1^58–61. In the zebrafish genome, Ivns1abp has been duplicated into ivns1abpa (GenBank NM_199279) and ivns1abpb (GenBank NM_201483), which are situated on chromosome 20 and 2 respectively. The spatio-temporal expression of both genes was analyzed by RT q-PCR and wholemount in situ hybridization. Ivns1abpb was highly expressed in zygotes indicating maternal deposition in the egg, while ivns1abpa was expressed from 3-somites stage onwards (Figures S5A and S5B). While phylogeny and synteny suggest that ivns1abpa is closer to the human IVNS1ABP locus (Figure S5C), we knocked out both genes using CRISPR-Cas9. We identified, selected, and outbred independent germline frameshift alleles for each gene: ivns1abpa^7bpdel: c.195delGGAACCT; p.(Glu66Metfs*5) and ivns1abpb^{2bpdel-1bpins}: c.205delGGinsT; p.(Gly69Leufs*27) (Figure S6A). Each mutated fish line was bred to homozygosity and intercrossed allowing to generate double maternal zygotic ivns1abpa/b null fish (referred to as MZ dKO) (Figure S6B). QPCR and western blot analyses of MZ dKO embryos indicated that both transcripts were targeted for nonsense mediated decay, resulting in ivns1abpa/b protein-null fish (Figures 5A, 5B and S6C). MZ dKO fish did not display overt anatomical differences during early embryogenesis but exhibited a shortened lifespan with approximately 45% dying within the first month post fertilization. If bred in isolation, a few MZ dKO escapees could be grown up to one year of age (Figure 5C). Compared to WT, MZ dKO fish had a propensity to swim less and remained at the bottom of tanks. Based on the peripheral neuropathy diagnosed in the three IVNS1ABP^F253C/F253C patients and the cellular defects observed in patient-derived MNs, we next assessed the development of primary MNs in zebrafish larvae. During the first day of development, neuromuscular connections are limited to three primary MNs per hemisegment, including caudal primary (CaP) MN which is the only one extending an axon toward skeletal muscles, leading to spontaneous motor activity^62,63. We therefore investigated the morphology of CaP MNs at 32 hours post fertilization (hpf) using an anti-Synaptotagmin 2 antibody⁶⁴. Microscopic observation of trunk hemisegments between somites 2-7 revealed a significant increase of poorly fasciculated CaP MNs in MZ dKO (90%) compared to control (28%). Aberrant CaP axon pathfinding and ectopic branching in MZ dKO (35%) compared to control (5%) was also readily documented and quantified (Figure 5D). These abnormal primary MN phenotypes evoke those seen in gan knockout fish⁵⁹, indicating that ivns1abp may equally be required for proper axonal development of MNs.

• Download figure
• Open in new tab

Figure 5.

Maternal zygotic double knockout of ivns1abpa and ivns1abpb leads to defective motor neuron development in zebrafish

(A) qRT-PCR of maternal zygotic double knockout (MZ dKO) embryos at 24 hours post- fertilization (hpf) shows depletion of both ivns1abpa and ivns1abpb transcripts. See also Figure S6A. Unpaired t test, *** p < 0.001, n=3 biological samples, with at least 20 embryos for each biological sample.

(B) Western blot of MZ dKO embryos at 1-day post-fertilization (dpf) shows absence of ivns1abp proteins. See also Figure S6B.

(C) Representative image of 1-year-old WT and MZ dKO female fish shows no obvious phenotypic difference. Scale bar, 1000 µm.

(D) Representative confocal z-stack images of 4 hemisegments of 31 hpf WT and MZ dKO embryos immunostained with Znp1, as well as a schematic representation of the rostral, middle and caudal primary MNs (RoP, MiP and CaP respectively) with their axon trajectories in each hemisegment (A: anterior, P: posterior, D: dorsal, V: ventral). Compared to normal CaP motor neuron axons, MZ dKO embryos show higher % of poorly fasciculated ventral axons (*** p < 0.001) and aberrant pathfinding (**** p < 0.0001), with mild to severe projections that branched out.

(E) Exemplary swimming paths extracted from 2-minute videos of three WT and three MZ dKO 7 dpf larvae, with each larvae in individual well of a 24-well plate. WT larvae exhibited higher swimming distance and speed compared to MZ dKO, especially after 10 minutes exposure of 5 mM PTZ. Colors represent different velocity ranges, with blue: low (< 8mm/sec), yellow: intermediate (8-16 mm/sec), and orange: high (> 16mm/sec). x and y Axis represent pixels.

(F) Compared to WT (n=72), 7 dpf MZ dKO (n=67) larvae show significant differences in locomotor behaviors, including shorter distance swam, higher total pause time between swim bouts, and lower high-speed bouts where velocity was at least 16 mm/s. Larvae were challenged with 5 mM of Pentylenetetrazol (PTZ) for 5, 10, and 15 minutes (mins). In Gardener-Altman and Cummings estimation plots, the mean difference between the two groups (black dot) and 95% confidence intervals (black bar) are plotted on the secondary axis. The orange shaded area in orange shows the 5000 bootstrapped confidence intervals. Two-sided permutation t-test for p-values are represented as * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001. See also Figure S6D.

Since MNs extend their axons toward skeletal muscles, leading to spontaneous motor activity, we next assessed whether these developmental anomalies could impact fish motility. At 7 days post fertilization (dpf), MZ dKO larvae (n=67) showed weaker locomotor capabilities, characterized by significantly lower distance swam and longer pause times compared to WT larvae (n=72) (Figures 5E, 5F and S6D). These differences were even more pronounced at 5, 10 and 15 minutes post Pentylenetetrazol (PTZ) treatments, suggesting that at baseline and upon sensitization, MZ dKO fish exhibited impaired locomotion and overall swimming behaviors. These results are consistent with previous findings where defective axon pathfinding and axon over-branching in ccdc80-l1 or gan knocked-down zebrafish embryos led to motility issues⁶⁴. Altogether, these in vivo results obtained in a surrogate animal model revealed that the physiological function of ivns1abp towards CaP MN development is conserved between mammals and fish. The complete inactivation of both ivns1abpa/b genes in zebrafish complements and supports the notion that the identified missense variant in humans is behaving as a typical loss-of-function allele.

Are IVNS1ABP and gigaxonin functionally redundant?

The protein structure homology, the overlapping clinical presentations and shared protein substrates, point to a hitherto un-appreciated functional relatedness between IVNS1ABP and gigaxonin.

Our ubiquitome analysis revealed Tubulin-folding cofactor B (TBCB) as the most differently ubiquitinated in patients’ cells compared to healthy controls. TBCB was shown to interact with gigaxonin for its targeted degradation via the UPP⁵⁰. TBCB is well known to play a role in neuronal growth cones, controlling microtubule dynamics and plasticity⁶⁹. Whereas TBCB knockdown promotes axon growth, TBCB overexpression leads to microtubule depolymerization and axonal damage, which are witnessed in GAN1 patients⁷⁰. Vimentin, Peroxiredoxin-1 (PRDX1) and 14-3- 3 proteins are additional IVNS1ABP targets that overlap with previously published gigaxonin pull- down experiments^46,51. For instance, Vimentin accumulation in gigaxonin-deficient patients’ fibroblasts⁴⁷, a tell-tale sign of GAN1, was readily observed in IVNS1ABP patient’s cells as well. It is worth mentioning that the aggregation of mutant Vimentin in a patient carrying a heterozygous p.(Leu387Pro) variant is also associated with a progeroid syndrome with motor and sensitive peripheral neuropathy⁷¹. This suggests that the accumulation of Vimentin might be an important driver of the pathogenesis of IVNS1ABP deficiency. Similarly, the dysregulation of the antioxidant PRDX1 enzyme, known to control motor neuron differentiation⁷², and 14-3-3 proteins which interact with neurofilament protein-L and regulate dynamic assembly of neurofilaments⁷³ were further evidence that IVNS1ABP and gigaxonin converge on a similar set of substrates.

MAP1B is another well-established substrate of gigaxonin. Mutations in the KELCH domain of gigaxonin were shown to disrupt its interaction with MAP1B, leading to its accumulation and abnormal elongation and guidance of axons⁷⁴. Upon synthesis, MAP1B full length (FL) is cleaved in MAP1B heavy chain (MAP1B-HC) which possesses a microtubule binding site (MTB) at its N- terminus, and MAP1B-LC which contains an actin binding site in its C-terminus^74,75. Although MAP1B was not part of the top 10 targets in our ubiquitome analysis, the Ubiquitin-conjugating enzyme E2 L3 (UBE2L3), a known binding partner of MAP1B-LC that favors UBE2L3-mediated UPP⁵², significantly accumulated in patients’ NPCs compared to control cells. Accordingly, we found that endogenous MAP1B (FL, HC and LC) protein levels across different cell types significantly accumulated in IVNS1ABP-mutated and -null cells, as they did in GAN-null patient’s fibroblasts. Our results documenting similar MAP1B and Vimentin accumulation in IVNS1ABP and gigaxonin single knockout fibroblasts argue in favor of both being needed for their effective turnover. However, one may be surprised that double IVNS1ABP/GAN knockout cells do not show worsened MAP1B levels than what is seen in single KOs. This indicates that IVNS1ABP alone cannot compensate for the loss of gigaxonin and vice versa. This is supported at the organismal level in humans or zebrafish, where the inactivation of each gene produces a phenotype that is not entirely masked by the remaining active paralogue. To examine the full extent of a possible functional redundancy between IVNS1ABP and gigaxonin, triple KO zebrafish or double KO mice will need to be generated to better appreciate if compensatory mechanisms take place, which would speak to the need of having both active paralogues to enforce tight proteostasis. Notwithstanding, our results indicate IVNS1ABP and gigaxonin are responsible for protein clearance of common substrates, which is reflected in the strong overlapping symptomology seen in both syndromes.

The moonlighting function of IVNS1ABP

Besides controlling proteostasis, IVNS1ABP seems to play different roles in multiple biological pathways, including spliceosome machinery^35,76,77, signal transduction⁷⁸, c-Myc transcriptional regulation⁷⁹, and actin stabilization and cytoskeleton organization⁸⁰. Concurrent investigation by Fang et al., (2024) using the same isogenic patient cell lines carrying homozygous WT, p.F253C or KO alleles of IVNS1ABP showed that this E3 ligase is also regulating actin dynamics. For instance, IVNS1ABP^F253C/F253C-mutated or IVNS1ABP^KO/KO cells displayed dysregulation of actin polymerisation, which led to impaired cytokinesis and mitosis, resulting in cell cycle arrest and premature senescence. All in all, our parallel investigations have revealed IVNS1ABP’s function towards maintaining cellular homeostasis by controlling the levels of key cytoskeletal proteins.

Therapeutic intervention

Owing to the molecular, cellular and clinical signatures shared by GAN- and IVNS1ABP-deficient patients, we suspect that similar modalities could be employed for therapeutic interventions. Recently a small molecule screen in morpholino-depleted and knockout gan zebrafish has identified a few hit compounds⁵⁹. These small molecules belong to the class of muscarinic antagonist and α-adrenergic agonist and proved to be efficacious in stabilizing the neuro- muscular junction and increasing the locomotion of gan deficient fish. Future experiments will need to assess whether these molecules can also prove to be effective in rescuing the locomotion in double ivns1abpa/b mutant fish, which would lend further credence to the notion that both diseases share a common pathogenesis.

The investigational gene therapy led by the NIH involving the single injection of a modified scAAV9/JeT-GAN virus into the fluid surrounding the spine of GAN1 children is an alternative strategy to restore adequate levels of gigaxonin⁸¹. The promising observation of a dose- dependent slowing down of motor decline in treated children suggest that similar approaches could be envisaged for IVNS1ABP-deficient patients which may also benefit from an exogenous supply of gigaxonin.